CALL (812) 333-8888 FOR MORE INFORMATION OR E-MAIL info@copa.org
| |||||||||||||||
CALL (812) 333-8888 FOR MORE INFORMATION OR E-MAIL info@copa.org | |||||||||||||||
 |
|||||||||||||||
Reports: EPA | |||||||||||||||
Dioxin Reassesment ReviewU.S. Environmental AgencyScience Advisory Board Section 4 4. DETAILED FINDINGS--HEALTH DOCUMENT 4.1 Disposition and Pharmacokinetics Issues 4.1.1 Strength of the database (Charge Question 2) The Charge for this review (see section 2.2) states that An understanding of the relationship between exposure and dose is an important aspect of an adequate characteriza tion of risk. A specific concern is how the extant data are used to predict tissue dose levels of 2,3,7,8-TCDD in humans under low exposure conditions. There is a an extensive animal data base relating exposure to tissue dose for 2,3,7,8- TCDD (although data are lacking for many of the related compounds), and there is a substantial (and generally solid) body of data on the absorption and distribution of 2,3,7,8- TCDD in animals (Birnbaum, 1985; Gasiewicz et al., 1983; Neal et al., 1982; Olson et al., 1983; Van den Berg et al., 1994). There are insufficient human data to support deposition and tissue dose modeling however, and this data gap severely restricts animal (largely rodent) to human extrapolation. Further, there are only a limited number of animal studies that reflect accurately likely environmental exposure scenarios for humans. For example, gastroin testinal absorption of 2,3,7,8-TCDD is influenced by the presence of other compounds, the nature of the matrix, and the very limited aqueous solubility of 2,3,7,8-TCDD. These and other factors relevant to the partitioning of 2,3,7,8-TCDD and gut absorption are of critical concern, given the fact that foodstuffs are considered by EPA to represent a major source of chronic low level human exposure to dioxins . The pulmonary exposure data base is based on the pulmonary absorption of 2,3,7,8- TCDD from solution; absorption by this route is known to be high. The most likely non- dietary human exposure to 2,3,7,8-TCDD, however, would be through inhalation of particu late matter incorporating dioxin in a solid matrix (e.g., fly ash). Although data on pulmonary absorption from a solid matrix are available (Nessel et al., 1990; Nessel et al., 1992a; Nessel et al., 1992b), these data are not addressed in the reassessment document. Given the large data base, a more thorough analysis of the biological determinants of tissue absorption and deposition (particularly with low dose exposure) should be carried out. For example, mice show a different pattern of distribution to the skin than rats, with the former more accurately reflecting data obtained in non-human primates (Brewster and Birnbaum, 1988; Gallo et al., 1992; Rahman et al., 1992). The hepatic distribution patterns of 2,3,7,8-TCDD in animals appears to be dose- dependent and saturable. This is an area that needs further study, particularly with regard to potential interspecies differences and the development of valid human physiologically-based pharmacokinetic (PBPK) models. In general, the reassessment document has drawn extensively from the existing animal data relating exposure/tissue dose and absorption/distribution, but did not characterize adequately the strengths and weaknesses of the data- base. Further, the inferences drawn leave some issues unaddressed. A greater effort needs to be made to describe the long-term effects of the decrease of PCDD- and PCDF levels in the environment, and in turn, in human exposures. This description should provide an improved estimate of the mix of chemicals in the TEQ and their capacity to be both agonists and antagonists (or synergists) in the overall biological effects of 2,3,7,8- TCDD and 2,3,7,8-TCDD-like compounds, as well as those chlorinated compounds that are reviewed in the health assessment documents but are not 2,3,7,8-TCDD-like. In addition, to enhance further the value of the reassessment, a greater effort is needed to provide a better understanding of the consequences and possible interaction of employing both the receptor model for risk assessment and the utilization of the concept of TEQ. For example, what can be expected in this context with a range of halogenated chemicals that may interact in various ways as either agonists or antagonists? What can be expected from a mixture of such compounds, when considering both additive or non-additive effects, with constant or changing concentrations of those individual components which contribute to both response and the TEQ value? In chapters I through 7, which mainly review the current research data, the document provides a generally balanced evaluation, and inferences from these chapters are appropriate except as noted above. However, Chapter 8 presents an incomplete review of the dose response model (Linear Multi-Stage [LMS] and modified LMS) approach, with no consider ation of alternative models used by other agencies, nor discussion of the literature. This is reflected in the summary chapter. Thus, the Committee recommends that EPA provide either additional discussion of alternative approaches and their implications for risk assessment in Chapter 8, or present a clear justification for choosing this particular dose-response approach over others. Chapter 9 should also be modified to reflect these additions. 4.1.2 Disposition and Pharmacokinetics (Charge Question 3) In addition to the specific questions asked about disposition and pharmacokinetics by the Charge, the introductory sentence leading into this issue must also be evaluated. This sentence states The evaluation of available data and the development of physiologically- based models has led to a better understanding of the disposition and pharmacokinetics of dioxin and related compounds than for most other environmental chemicals. This sentence implies that these models are well-established and ready for incorporation into the current assessment. It is the Committee s consensus, however, that there are several issues that should be addressed before these models can be used effectively in the risk assessment. The following questions should be addressed in a revised assessment document: a) Does the database to determine mechanism and models apply only to 2,3,7,8-TCDD or may it be extended to other related compounds? Currently the great majority of data on physiologically based kinetic models is derived from 2,3,7,8-TCDD research, but about 90% of estimated risk is from related compounds (as stated in the reassessment document, p. 9-81). b) Do the models proposed extend to all dose ranges, particularly to low dose? The data used in current models were not generally derived from studies conducted at low dose levels. There is significant controversy as to the shape of the curve in the models proposed at the low dose levels. c) Are all organs effectively addressed in the models? The models appear to focus on liver and fat concentrations in rodents; however, the target organs in the epidemiology studies for carcinogenesis are the gastrointestinal tract and others. It is not clear that the models effectively describe the exposure and metabolism of those possibly impacted tissues. Particular questions may exist with respect to the lung for carcinogen exposures that occur by the respiratory route (although this route is probably of minor significance for most of the population, it could be of concern for occupational exposures). d) If models have primarily been determined from short-term or single-dose studies, do these results apply to chronic studies and/or longer term exposures? In particular, the chronic Kociba et al. (1978, 1979) studies, which included low exposure levels, did not necessarily fit one of the models proposed, whereas the dose finding studies did. A question then exists concerning the model s efficacy for the real life-low exposure situation. e) Do the models apply across relevant mammalian species? f) What is the relationship between dose as used in the dose-response models and tissue or body burden that the EPA uses in many comparisons between human and animal exposures and risks? Lastly, the variability of the half-life of 2,3,7,8-TCDD in the human leads to questions about the precision of the models. The estimated half-life discussed varied from five to 11.3 years. The range given here may represent a difficult communication issue, and affects the precision of the model. In the body of the health reassessment document, the issue of PBPK modeling is discussed in three chapters (1,8, and 9). However, there is little evidence of actual integra tion of any of these models into specific portions of the document. The concept of physiologic-based modeling is discussed separately in the risk assessment portions of the document, and is conceptually a part of the process. Scrutiny of the document (and oral discussions with the document s authors at the public meeting) however, yielded no evidence of use of a specific PBPK model in the risk assessment. Given the issues noted above, this is understandable and not considered to be a particular weakness by the Committee (however, opportunities do exist--see the discussion below), but future revisions of the document should be clear as to the degree to which such modeling has been incorporated into the assessment process. Notwithstanding the lack of incorporation of these models into the current reassess ment and the significant questions noted above, the database on dioxin could provide an opportunity to utilize state-of-the-art PBPK models. The Carrier-Brunet-Brodeur model (Carrier et al.,1995a; 1995b) (which was not yet published at the time the Health Reassess ment was under development), for example, incorporates non-linear elimination rates; as body burden declines, half-life increases. The applicability of this model and its implications should be discussed in any future revision of the reassessment document. The opportunity to utilize the Ah receptor binding in modeling is significant, and these models should be more effectively incorporated into the risk assessment through revisions to Chapter 8. 4.1.3 Incremental Exposures and Bioaccumulation (Charge Question 4) The EPA reassessment document provides a large amount of animal and human data regarding bioaccumulation of 2,3,7,8-TCDD. However, it is readily apparent that a large gap exists in our knowledge regarding the pharmacokinetics of common, environmentally occurring congeners of 2,3,7,8-TCDD. If Toxicity Equivalency Factors (TEF) are going to be used to assess human toxicity, and if approximately 10% of the total exposure to dioxins is attributable to 2,3,7,8-TCDD, an accurate estimation of total potential for toxicity can be made only if information is available regarding distribution, metabolism, and half-lives of other major contributing components (See discussion in section 4.13). It is not possible to make a general estimate of toxicity if we know nothing of the pharmacokinetics of the majority of the dioxin-like compounds to which humans are exposed. The reassessment is silent on this issue, but data from autopsy evaluations have been published by Schecter (1991) and should be used (if possible) to extend our knowledge of the pharmacokinetics of these compounds. Major shortcomings are evident in the data base and the relevant issues have not been dealt with adequately. For example, no attention has been given to reviewing the half-lives of many of the important compounds that constitute the TEQ. Data from studies concerning exposures of subsistence fishermen appear to be in conflict with the EPA estimation of body burden (Columbia River Intertribal Fish Commission, 1993; Dewailly et al., 1994; Svensson et al., 1991). These studies suggest that the reassessment's estimated body burdens may be as much as two orders-of-magnitude higher than actual levels; this difference may be related to errors introduced by not accounting for the various half-lives of the agents included in the TEQ. 4.1.4 Uncertainties in Back Extrapolations of Body Burden (Charge Question 5) The Committee s primary concern with the topic of the body burden back extrapola tion lies with EPA s treatment of uncertainty regarding the half life of dioxin, and the exposition of the methods for the back calculation in the health document. As discussed at the public meeting, the Committee anticipates that future revisions of the reassessment document will incorporate the latest data set from the Ranch Hand study, which should narrow considerably the range of estimates for dioxin half-life, and so reduce the uncertainty (from that source) in the back extrapolation of body burden. In addition, we expect that EPA will revise the discussion on the back calculation method to include material covered in Chapter 6 of the exposure document, but not currently addressed in the health document. 4.2 Mechanisms of Dioxin Action 4.2.1 Animal-to-Human Extrapolations of Receptor Structure and Function (Charge Question 6) Mechanisms of dioxin action are discussed in various parts of the Health Assessment document. Overall, the presentations are very useful. Whether or not the document addresses all the relevant information and alternatives depends, however, on the section that is being read. Chapter 2 offers an unequivocal assessment that ...the Ah receptor mediates the biological effects of TCDD. Yet, when reading Chapters 3, 4 and 5 dealing with specific toxic events, it becomes clear that numerous fundamental uncertainties occur and mecha nisms of action for the toxic events beyond receptor binding are largely unknown. In Chapter 9, however, there is a return to the acceptance presented in Chapter 2. This may indicate that the authors of Chapter 9 are willing to go along with the Ah receptor information as it exists today. One major difficulty lies with the use of the term "mediator" and perhaps confusion about what constitutes a mechanism of action. Other than what is described in the sections dealing with developmental toxicity (e.g., cleft palate, hydronephrosis), very little is known about the biological steps that finally lead to frank toxicity; the lack of information on immunotoxicity is particularly limiting. Much of what is purported to link the Ah receptor to specific toxic events is merely the demonstration of an association between the binding of TCDD to the receptor and an eventual appearance of an adverse effect some time later in some species (when the interspecies variation in doses that produce lethality can be over 8,000-fold, based on the data provided in Table 3-1, Volume 1, of the reassessment docu ment). But the possible downstream events, if they exist, between Ah receptor binding and the final toxic manifestation are not well established. Mechanism of action should mean that at least some of the intermediate steps, after Ah receptor binding and leading to the pathologic processes involved, are known to some extent. The loose use of the term "media tor" implies that the association apparently observed is in reality the initiator of the process. In actual fact, the only mechanism of action involving the Ah receptor that has been worked out sufficiently well to be called the biological sequence that describes a "mechanism of action' is the induction of cytochrome P450. What is known about the induction process is truly elegant. The rest of the biological consequences of TCDD exposure are yet to be described adequately and sequentially in mechanistic terms. The document (Chapter 2) presents an excellent review of what is known about the Ah receptor and the multiple steps involved in TCDD-induced cytochrome P450 induction. Research findings during the last 10 or so years that have identified the structure and mode of action of the Ah receptor represent a major scientific achievement. How a planar polar compound such as TCDD binds with a cellular receptor, followed by translocation into the nucleus, where transcriptional activity of DNA is influenced, is very well understood. On the other hand, there is a large intellectual chasm between the elegant science describing the details of the TCDD receptor and its mechanism of initiating a cellular response, and the poorly understood manifestations of the toxic events associated with an alteration of the homeostasis of an animal. The linkage between Ah receptor action and specific cellular toxicity remains undefined. Several Members of the Committee noted that the possibility that the Ah receptor system may be a sensing pathway to protect the cell is not consid ered, nor are there other attempts to put forth scientifically testable hypotheses. In any future revisions, EPA should present more clearly the major deficiencies that exist in the current mechanism database and provide some discussion of any plausible alternative hypotheses. Although the Ah receptor is likely involved in producing TCDD toxic effects of potential concern to humans, there are multiple levels of regulation of the receptor pathway. The induction of CYP1A1 does not serve as a good model for all receptor- mediated responses to dioxin, particularly those that result in altered patterns of growth and differentiation. The studies of Poland and Knutson (1982b) in hairless mice indicated that for responses such as epidermal hyperkeratinization and skin tumor promoting activity, the Ah receptor is necessary, but not sufficient. Two implications from these studies are: a) that toxicity is under multiple genetic control, and b) the most sensitive response in animal models is not necessarily the most valid predictor of toxicity in humans. Chloracne remains the most definitive response documented in humans and clearly occurs at high exposure levels. However, there are relatively few animal models for chloracne. Most of the mechanistic data support the involvement of the Ah receptor, but say little (in the context of toxicity), about how the activation of this protein alters normal physiologic function and/or development. Risk assessments based solely on Ah receptor activation or on the existing knowledge of CYP1A1 induction are unlikely to provide a biologically defensible prediction (quantitatively or qualitatively) of likely toxic outcomes in humans, particularly under low exposure scenarios. Regarding the Ah receptor in humans, there is adequate evidence that, in overall function, the human Ah receptor mechanism essentially acts the same as the Ah receptor in rodents and in other laboratory species. The Ah receptor in humans, however, has an affinity for TCDD that is lower than the affinity in C57BL/6 mice or in most laboratory rat strains. There may be at least a 10-fold range of variation among the human population in the affinity with which the Ah receptor binds TCDD. Induction of CYP1A1 exhibits classical sigmoidal log-dose response curves in several human cell lines in culture. The EPA document s phrase ...and the role that the purported mechanisms of action might contribute to the diversity of biological response seen in animals and, to some extent, in humans? in this question can be better posed as ...how convincing is the evidence for the purported mechanisms that link receptor binding to toxic effects in humans? Unfortunately, the evidence is quite mixed. There is only limited evidence for toxic effects in the Ranch Hand study (USAF, 1991; CDC, 1988a,b). Studies of the Seveso population (Bertazzi et al., 1993; Bertazzi et al., 1989; Mocarelli et al., 1986) show significant excesses of multiple myeloma and hepatobilliary tract cancer in women, and lymphoreticulosarcoma in men in zone B, and soft tissue sarcoma in men in zone R. These lesions, however, were seen in the zones of lower, not higher, estimated exposure. Conversely, and further clouding the issue, these same studies also report decreases in breast [Some Committee Members suggest that this reduction may result from the anti-estrogenic effects of dioxin.] and other cancers in females (although the number of cases is very small and the decrease may be spurious). In the exposed chemical plant workers studied by the National Institute of Occupational Safety and Health (NIOSH) (Fingerhut et al., 1991), there is no excess cancer of any kind in the less heavily exposed workers (i.e., those who were exposed for less than one year). Workers with over one year s exposure showed statistically significant increases in respiratory system cancer, but they were exposed to a wide variety of potentially carcinogenic agents in addition to dioxin. Given the possible confound ing, and the somewhat equivocal links of dioxin to excess cancer in the group as a whole, it is difficult to document a dioxin-cancer relationship. [Several Members of the Committee suggest that EPA has neglected other human exposure studies of interest, particularly the health effects seen in the Taiwan rice oil PCB poisoning episode (as reported in Rogan et al., 1988; Wu et al., 1984; Chang et al., 1982a; Chang et al., 1982b: and Chang et al., 1981) which could reduce the reliance on animal-to-man extrapolation.] [Noting that cancer excesses were seen in the Fingerhut et al. (1990; 1991) studies only in workers with 20 or more years of latency, one Committee Member argues that the Bertazzi study should be discounted as uninformative. It was carried out only ten years after exposure, the levels of ex-posure to the people in the zones that are characterized as having excess cancers are far less than those of the workers studied by Fingerhut et al., In this Member's opinion, both the reported excesses and deficits are more likely to result from chance than from any dioxin effect.] 4.3 Toxic Effects of Dioxin 4.3.1 Animal Models for estimating Human Risk (Charge Question 7) The Committee believes that the EPA reassessment document reviewed and summarized adequately the current knowledge base on the experimental disease(s) produced by TCDD and related compounds in animals. The Committee agreed that, because there is a limited amount of human data, EPA will have to rely on the results of animal studies for some portion of its risk characterization. However, the Committee also finds that it is probably not appropriate for the Agency to single out the results of a given study (unless it is of a seminal nature) for decision making. The EPA would be better served to employ a weight-of-the-evidence approach, using the totality of the data. The Committee noted that although there were many interspecies consistencies in response following exposure to the compounds in question, there were also many signifi cant inconsistencies. The Committee questions whether some of the interspecies differ ences were true differences or were a reflection of the dose levels used in the different studies. A portion of the interspecies Inconsistencies may be attributed to the fact that some of the animal studies involved lethal exposure levels but that other studies did not. For example, some of the effects noted in animals (see Table 9-2 of the reassessment document), e.g. wasting disease, chloracne, testicular atrophy, hepatotoxicity, cardiovas cular lesions, hypoglycemia, edema, and porphyria, were found primarily in animals that died. Many of these effects were not observed in cows, however, and none of the cattle studies involved lethal exposure levels. Therefore, the lack of interspecies comparability may indeed be a reflection of the dose levels used in particular species/studies. Another apparent inconsistency which may be a reflection of study design is related to the carcinogenicity of TCDD. Although TCDD has not been reported to be carcino genic in guinea pigs, rabbits, chickens, cows, or monkeys, none of the studies were of sufficient length to ascertain whether it is indeed carcinogenic or not in these species. Additionally, the lack of observed chloracne in a given species may be related to anatomi cal differences; many of the tested species have "fur rather than true "hair follicles as are found in monkeys and humans. Also, a feather follicle (chicken) is anatomically quite different than a hair follicle. Some discussion of these points would be beneficial to readers of the reassessment document. The Committee also suggests that the document discuss "primary" versus "secondary" effects of exposure. For example, some of the effects related to TCDD exposure (testicular atrophy, cardiovascular pathology, edema, and porphyria) may not be directly related to the compound, but may rather be a reflection of the sick animal syndrome. Comparative mechanistic and pharmacokinetic data would be needed before making a conclusion that a given effect is directly attributable to TCDD or related compounds. The degree of uncer tainty in this regard needs to be discussed in the document and needs to be reflected by using ? marks in Table 9-2. The Committee believes that Table 9-2 is extremely important and is very likely to be used by the casual reader; consequently it should be made as accurate as possible. Members of the Committee have noted errors in the Table, and we suggest that the tabular material in the document be reviewed for accuracy. In summary, the Committee finds that there is evidence of both inter-species consistency and inconsistency in the EPA document. These variances need to be addressed in an objective manner in the document, and discussed in conjunction with all the uncertainties inherent in the various endpoints. In addition, the Committee felt strongly that an examination of the totality of the animal data will be required in the final hazard characterization. Finally, nearly all of the Members of the Committee take strong exception to the definitive sentence on page 9-78 (see also the Committee s discussion of Charge Question 1, in section 2.2.2 of this report) of the EPA document, which states: The scientific community has identified and described a series of common biological steps that are necessary for most If not all of the observed effects of dioxin and related compounds in vertebrates, including humans. Binding of dioxin-like compounds to the cellular protein Ah receptor represents the first step in the series of events attributable to exposure to dioxin- like compounds, including biochemical, cellular, and tissue-level changes in normal biological processes. Binding to the Ah receptor appears to be necessary for all well studied effects of dioxin but is not sufficient, in and of Itself, to elicit these responses. This pronouncement is too strong. Virtually all the Committee believes that it is more accurate to state that binding of TCDD and related compounds to the Ah receptor is a marker of exposure, but has not yet been established to be necessary for the induction of several of the observed effects. This degree of uncertainty needs to be stated in the document. 4.3.2 Variations in Human Sensitivity (Charge Question 8) The issue of human sensitivity may be divided into two questions. First, are human sensitivities so distributed that a representative average can be assumed? Secondly, do humans on average yield a response which might be considered to be average relative to the spectrum of responses in animals? Wide variations in response to dioxins are well documented in animal studies, with at least a three-fold order-of-magnitude being reported for some responses between animal species and even within a given animal species such as rats and mice, or between very young hamsters and adult hamsters. Responses of humans are known to vary by several orders-ofmagnitude with respect to the exposure to many exogenous substances as drugs, where some individuals are known to be responders while others are considered non-responders. It is reasonable to assume, therefore, that responses of humans to TCDD and its congeners will vary widely. Furthermore, observations reported in human exposure studies as from Seveso, Italy (Bertazzi et al., 1993; Bertazzi et al., 1989; Mocarelli et al., 1986) and the U.S. Air Force Ranch Hand Study (USAF, 1991; CDC, 1988a,b) as well as other studies of occupational cohorts clearly indicate that wide variations in the frequency and severity of response occur. [One Member of the Committee objects to this statement, citing findings in this review which assert that the only adverse human health effect tied to dioxin is chloracne. Given this latter finding, he does not accept that studies clearly indicate that wide variations in the frequency and severity of response occur."] The Committee finds that there is no single animal model that could accurately predict human responses. EPA, in its revision, should identify clearly the limitations of existing animal models in terms of their ability to predict the various health outcomes that may occur in humans as a result of exposure to dioxin and dioxin-like compounds. Are humans expected to yield an average response relative to the spectrum of responses in animals? Humans could be as sensitive as the most sensitive mouse species, or insensitive as the least sensitive mouse species to TCDD. Based on the available human data, it is debatable whether the most sensitive species, or the most representative animal species, should be used when selecting an animal model to predict TCDD toxicity in humans. Ideally, if a high degree of confidence in the model existed, one should use the animal species that is most representative of humans. If no single model is appropriate, animal models should be selected that permit a conservative approach to be employed with respect to extrapolation to human subjects. What is unclear with respect to the question regarding average sensitivity is what constitutes an average response in humans who exhibit an average sensitivity, and at what level of exposure, both acute and chronic, does this occur? When considering toxicity to humans, one must always consider the most sensitive population. Can highly sensitive sub-populations be identified that are at greatest risk to dioxin exposure? Such sub-populations might include pregnant women, infants, and children or members of a population with an above average exposure as populations whose subsistence diet consists largely of fish. Although variations in response are reported and presented in the document, the document (in the opinion of most of the Committee) does not (and perhaps cannot, given that dioxin-specific effects beyond chloracne--see the discussion below--are not established) concisely articulate that wide variations in human sensitivity to dioxin exposure occur and should be anticipated. The emphasis is heavily placed on low level exposures that might cause toxicity in some individuals. [Several Members have noted that they find the EPA emphasis on low-level exposures to be appropriate and consistent with prudent public health practice.] 4.4 Chloracne as an Indicator of Exposure (Charge Question 9) The Committee believes that the EPA document reflected adequately the current knowledge base on chloracne as it relates to the subject compounds. Chloracne is a clear indicator of exposure, but the absence of chloracne in an exposed subject is not an indicator of low exposure. In fact, the Committee s consensus is that chloracne is the only lesion of note clearly established as being related to TCDD exposure; in the absence of sufficient data on human tissue levels at peak develop ment of chloracne, however, a dose-response relationship is difficult to ascertain. [Several Members of the Committee believe that recent research findings published after the Committee s public meeting (e.g. Kogevinas et al. 1995; Huisman et al., 1995) establish some cancer and developmental effects in humans as outcomes of either CDD or TCDD exposures. ] The Committee also noted that chloracne has also been found in people exposed to related compounds such as dibenzofurans and PCBs. 4.5 Cancer 4.5.1 Epidemiological Evidence (Charge Question 10) When a difference between compared groups is observed, causation may still not be imputed. Similarly, when no difference is seen, it cannot be concluded that the study variable is not associated (noting, of course, that the presence of association does not impute causation) with some outcome or is not causal. Causation is not itself an experi mental or epidemiological result but a judgment made about the results. In making such a judgment an epidemiologist takes into account the possibility that bias and chance may play a part, but additionally examines the observed association in terms of certain characteristics associated with "causal" relationships. These are sometimes referred to as the Hill criteria (for A. Bradford Hill, who first codified them). They are: a) strength of the association; b) consistency; c) specificity; d) relationship with time; e) biological gradient (dose-response relationship); f) biological plausibility; g) coherence of the evidence; h) observed change following some intervention; and i) analogous findings. The reassessment document mentions three of the Hill criteria (gradient, consis tency, and strength) but does not explicitly use them as a tool to organize its discussion of causality. The Committee does not regard this as a failing, since the Hill criteria are not so much rules, as viewpoints to aid in interpretation. The EPA response is that causation judgments were not explicitly part of Chapter 7, although the Committee notes conclusions of Chapter 7 were used as a partial basis for Chapter 9. We find this acceptable, since the entire discussion is couched in terms of characteristics of causal associations. Understanding the operation of bias and chance is especially important in interpret ing so called "negative studies" (studies where no differences are apparent, or where the differences are not statistically significant ). Differences produced by real effects can easily be masked by poor exposure classifications (misclassification bias), Chance can appear as a possible explanation merely by virtue of a small population available for study (poor statistical power), and potential risks can be undetectable by observing the exposed population for too short a time (bias produced by failure to account for adequate latency), to name just a few factors complicating interpretation of such outcomes. On the other hand, factors that can produce spurious increases in exposed groups in environmental epidemiological studies are much less common, most forces operating to lower the observed risks, not raise them. The reassessment document does a quite good job of taking these limitations into account, but could still benefit from additional discussion of the effects of confounding factors. Specifically, EPA should incorporate in a revised Chapter 9 the means of addressing confounding factors discussed by Agency staff at the review meeting. To summarize the above discussion, evaluating internal validity requires the assessment of the roles played by bias, chance and real effect. Each can operate, some times reinforcing other factors, sometimes offsetting other factors. There is often disagree ment about studies among experts, stemming from differing weights each places on the influence of bias, chance and real effect. Such differences in science are common, both in and out of the regulatory process. The Assessment is explicit about the judgments it makes, allowing others to differ if they feel that other emphases are warranted. The Committee feels this is preferable to merely cataloging variant points of view. An evaluation of internal validity helps a scientist in deciding how much to rely on the specific result of the experiment or study. It does not tell a scientist how much to extend that result to contexts or situations different than the one studied, i.e., how much to generalize the result. A separate evaluation for external validity is needed. The limits and extent of generalization are given by a study's external validity. In our context the question is whether a proposition developed in one context (e.g., a high dose occupational study like that of Fingerhut et al., 1990) can be generalized to cover other contexts (e.g., environmental exposures). Unfortunately, there are no rules for how far to generalize, if at all. Each study must be evaluated in a specific context to determine the extent to which it can be generalized. Cross-species generalization is not a problem for the epidemiological studies, but the question of generalizing to environmental doses remains. Interestingly, Chapter 7A of the Assessment (Epidemiological Data, Part A: Cancer Effects) does not comment on the applicability of the epidemiologic data to environmental exposures, satisfying itself with answering the question of whether TCDD and related compounds have the capacity to cause cancer in humans under any conditions. It answers in the affirmative, citing several studies of "sound design and adequate size" that have found a risk of soft tissue sarcoma (STS) (p. 7-74). Association of STS with dioxin exposure was raised by Hardell and colleagues (Hardell et al., 1979; Hardell, 1981a,b; Hardell and Eriksson, 1988; Hardell, 1993) and, according to the reassessment, has "stood up to extensive criticism and a great deal of subsequent research. (pp. 7-73, 7-74). The document also notes that the entirety of the association in these studies may be real and not due to selection bias, differential exposure misclassification, confounding, or chance." Although there are differing opinions about the validity of the Swedish studies, most Members of the Committee find that the Assessment clearly discusses the direction and degree of influence of the various sources of bias in these studies. The reassessment similarly discusses non-Hodgkin's Lymphoma (NHL) and, based upon a lack of a "minimally consistent picture of increased risks" fails to conclude at this time that dioxin exposure is related to NHL. This conclusion was arrived at by a clearly stated and scientifically defensible argument that is adequately supported. The reassessment also states that the epidemiologic data suggest that lung cancer might also be related to dioxin exposure, and cites the findings of the NIOSH study (Fingerhut et al., 1991) which reported statistically significant increases in respiratory system cancer (for those with over one year s exposure and a 20 year latency period. Because of possible uncontrolled confounding by smoking, as well as by exposure to many other carcinogens in the workplace, the EPA document is more tentative on this judgment, but considers that residual confounding is insufficient to explain the observed increase in respiratory system cancer risk. Here the basis for the EPA s position is weaker, reflecting the current lack of data on the possible nature of a dioxin/cancer relationship in the presence of confounding by smoking and occupational exposure to chemicals. Although the Committee does not reject the EPA s position, neither can it reject the alternative explanations which some Members of the Committee think have more merit. The document found insufficient data to make conclusions regarding stomach cancer, generally increased risk of all types of cancer, and sex differences in cancer risk. The Committee agrees. In summary, almost all Members of the Committee found that the reassessment's judgments on the epidemiology data (subject to the caveats noted) to be generally defensible. The document took into account many of the concerns of the broad scientific community and discussed them explicitly, but, of course, could not discuss all significant alternative view points. [Likewise, the Committee did not discuss in its public meeting, or address in this report, all the relevant epidemiologi cal studies noted in the reassessment document or the extant literature. ] The Committee does have some concerns about the ways and the extent to which uncertainties in the epidemiological data base were characterized. The reassessment docu ment (p. 7-77) refers to "uncertainties associated with the epidemiologic evidence." Some of these uncertainties are the usual ones attendant upon a subject where much research remains to be done and many questions are still unanswered. The more important uncertainties are those connected with the epidemiologic method itself. Observing some unintended or "natural" experiment in the real world, which is the essence of observational studies like epidemiology, has the enormous advantage that it involves human beings living under conditions similar to ones of concern to regulators and public health officials. For epidemiol ogy, the uncertainties are largely associated with the questions of internal validity extensively discussed in the Assessment itself (questions of bias, chance and real effect). The reassess ment document did not discuss remaining questions of external validity, the most important of which are the high exposure to low exposure generalization. The EPA should comment on this issue in any future revision, as well as on the relationships between agricultural and forestry, and environmental exposure levels (as well as varying exposure routes and patterns and associated environmental conditions and chemicals in these situations) and the cancers observed at those exposure levels. [One Member of the Committee notes that the reported effects of low-level exposure to forestry and agricultural workers predicts overwhelming incidence of certain tumors in the far more highly exposed production workers. Those tumors are not in excess or are barely in excess (see above) in the production workers. At least, this absence of external validation casts doubt on the general-ization of the forestry and agricultural studies. At worst, they indicate that those studies are flawed.] 4.5.2 Carcinogenicity of Dioxin-like Compounds (Charge Question 11) TCDD is one member of a large family of halogenated polycyclic aromatic com pounds. The EPA reassessment document describes such congeners and some of their metabolic and carcinogenic effects. A number of other studies have demonstrated the effectiveness of other dibenzodioxins and dibenzofurans, both individually and in mixtures, as promoting agents in the rat liver model (e.g. Nishizumi and Masuda, 1986; Waern et al., 1991; Shrenk et al., 1994). Although the data is as yet too sparse to make any extensive generalizations, it is reasonable to hypothesize that many of these congeners of TCDD will be effective promoting agents and thus carcinogenic at some dose. Another large class of closely related halogenated aromatic hydrocarbons are the polyhalogenated biphenyls. Many of the members of this class are promoting agents in the rat liver system (Sargent et al., 1991; Jensen and Sleight, 1986; Luebeck et al., 1991; Preston et al., 1985) and in other organ systems as well (Anderson et al., 1994). In addition many, but not all, polyhalogenated biphenyls interact specifically with the Ah receptor (Kafafi et al., 1993) at KD (the apparent equilibrium disassociation constant for ligand binding to the Ah receptor) levels that approach that of TCDD. However, the majority of the Committee concluded that the structure, metabolism, gene regulation and toxicities of this class, while overlapping with some of the characteristics of dibenzodioxins and dibenzofurans (Safe, 1990) are sufficiently different from those of the latter to argue that polyhalogenated biphenyls not be a part of this document. [Several Members disagree strongly with this finding and recommend the inclusion of polyhalogenated biphenyls in the assessment. ] They may be considered for future studies on assessments using this document as a model. Since many dibenzodioxins and dibenzofurans occur as components of mixtures, there have been several studies of the carcinogenicity of such mixtures indicating additivity of their promoting activity (Schrenk et al., 1994; Huff et al., 1991). In some instances synergy of components of mixtures of biphenyls have been reported (Sargent et al., 1991; 1992). Although the discovery and characterization of the Ah receptor were delineated using polycyclic hydrocarbons (halogenated or not) as ligands (Thorgeirsson and Nebert, 1977), it has been assumed that naturally occurring and/or endogenous ligands occur. Recent studies (Bjeldanes et al., 1991) identified some naturally occurring indoles as effective ligands. However, the contribution of such agents to the dioxin burden as ligands of the Ah receptor is unknown. Clearly more studies in this area are needed. In summary, dibenzodioxins and dibenzofurans which have been studied as congeners of TCDD, exhibit qualitatively similar toxicities, ligand reactivity with the Ah receptor, and show carcinogenic potential as promoting agents. However, the data in this field are too incomplete at this time to make generalizations in the direct application of findings with one member of the class, e.g., TCDD, to all others. Promoting activity of mixtures of this class can be additive and, in some cases, may be synergistic. The Committee (albeit with several Members taking exception) recommends that the polyhalogenated biphenyls, although having many similarities in their metabolic, toxic, and carcinogenic effects to TCDD, should be considered in a separate class and not considered in depth in this document. 4.5.3 Carcinogenic Activities of Dioxin and Dioxin-Like Compounds (Charge Question 12) The EPA reassessment document, in describing the experimental evidence, makes a clear distinction between studies that imply TCDD as a multi-site, complete carcinogen, as opposed to studies that emphasize the promoting properties of the agent. The principal animal assays that evaluated the carcinogenic potential of TCDD and some of its congeners are adequately reviewed. On occasion, however, emphasis is placed on the fact that TCDD is carcinogenic well below the maximum tolerated doses; what should also be noted is that quite often the response was statistically not significant (e.g., skin tumors in hamsters (Rao et al., 1988) and liver tumors in mice (Sugar et al., 1979)). In the document, the classification of TCDD as a complete carcinogen is done on an operational, not mechanistic basis. This is confusing since the classical definition of a complete carcinogen necessarily involves a consideration of the mechanism(s) whereby the agent effects its action. The term complete carcinogen has been reserved, until the advent of this and other recent documents on the carcinogenicity of TCDD (Huff et al., 1994), for agents capable of inducing all stages of cancer development, initiation, promotion, and progression (Boyland, 1980; Pitot, 1990). TCDD, as documented in the reassessment (as well as from other sources), is incapable of initiating cells in multiple in vitro and in vivo studies, and has never been satisfactorily demonstrated to have progressor activity. It should not be classified as a complete carcinogen any more than phenobarbital, phorbol esters, uracil or galactosamine would be considered as complete carcinogens. Thus, if the term complete carcinogen is to be retained as a classifica tion of the carcinogenic action of TCDD, a full definition of the term as used in the document must be given to prevent the confusion noted above. Furthermore, designation as a complete carcinogen implies in the minds of most readers direct mutational and clastogenic activity of the agent, which the evidence does not support for TCDD. The reassessment document also describes a second set of studies in which TCDD was characterized as having promoting capabilities. TCDD is classified as an ... extraordinarily strong promoter of liver and skin tumors. The following studies are referenced in the EPA document as support for this statement: a) Liver studies: Liver-tumor promotion by TCDD following initiating treatment with partial hepatectomy/DEN (Di-ethylnitrosamine)was first described in 1980 (Pitot et al., 1980). In female adult Charles River rats treated with a high dose of TCDD following partial hepatectomy, five out of seven eventually developed liver tumors vs. none out of four in controls. This was accompanied by an increase of enzyme-altered foci. No effects were seen in five animals treated with a low dose of TCDD. The findings are significant when a one-tailed Fisher s exact test is applied. In another study with adult female rats, a single dose of DEN was used as initiator and TCDD as a promoting agent (Graham et al, 1988). This study also suffers from low statistical power, and tumor data, at 60 weeks, are only available for one (the highest) out of three TCDD doses studied (5/8 vs. 1/5 in controls). According to the authors, "The number of animals was not sufficient to determine whether the incidence in DEN initiated/TCDD-promoted rats was different from that seen in rats treated with DEN alone." (A significant effect was claimed when DEN/TCDD-treated animals were compared to animals that had not received DEN - clearly an inappropriate comparison). A third study (referenced several times in the EPA document) was not published in the peer-reviewed literature (Clark et al., 1991) and the complete data do not appear to have been published elsewhere. The report should be specific in this regard. Data on tumor incidence are only given for intact and for ovariectornized animals initiated with DEN and treated with TCDD; apparently only one TCDD dose was used. No control data (rats initiated with DEN and treated with solvent) are presented. When the available data on liver tumor incidence are analyzed by Fisher's Exact test, they do not support the assertion that ovariectomy would provide a "protective effect" against liver tumor development. However, analysis of the same data with an uncorrected chi-square test, which is more appropriate for such an application (D Agostino et al., 1988) results in highly significant support (p<.01) for the hypothesis of a mitigation of effect in ovariectomized rats. Another study has become available in the open literature since preparation of the EPA reassessment document (Sills et al., 1994). The group sizes of animals used in this particular study provide better statistical power (between 12 and 15 animals per group) than do the previous studies. Despite the larger numbers, the increased tumor incidence in initiated-TCDD treated female weanling rats (five out of 15 animals) is not sufficiently high to demonstrate a difference from the controls (1 out of 12 animals) at a conventional level of statistical significance. It is interesting to note that, taken individually, two of the four studies on rat liver tumor promotion by TCDD cannot demonstrate a statistically significant effect, a point that should be addressed in the document. However, in all studies, there is evidence of a TCDD effect and, if the data from the four studies are pooled using appropriate statistical methodol ogies, the promoting effect of TCDD becomes highly significant. Furthermore, there is a substantial data base wherein, although no direct data on tumor incidence or multiplicity are provided, convincing evidence is given that treatment of initiated animals with TCDD increases significantly, and in a dose-dependent manner, the presence of altered hepatic foci and other signs suggestive of a strong promoting potential (Sills et al., 1994; Pitot et al., 1987; Dragan et al., 1992; Buchmann et al., 1994; Flodstrom et al., 1991; Waern et al., 1991). A recent and extensive study on the topic (Dragan et al., 1992) emphasizes the complexity of this particular endpoint. The EPA document describes several mechanistic studies and attempts are made to link the promoting activity of TCDD to such biochemical events as enzyme induction, internalization of EGFR, estrogen receptors, and similar endpoints. A recent study on inhibition of intercellular communication by TCDD might be added to this discussion (DeHaan et al., 1994). b) Skin studies: Promoting activity of TCDD was also shown in two skin studies: In 1982, TCDD was described as promoting skin tumor development in hairless mice (Poland et al., 1982a). The data presented in this paper fully support the assertion that TCDD is a skin tumor promoter. A second experiment (Hebert et al., 1990) claimed to confirm some of Poland's observations, but when read carefully actually did not quite do so. The table reporting the incidence of proliferative lesions does not indicate statistically significant observations. If the data are analyzed, only the number of mice with papillomas in the lowest TCDD group is statistically higher than in controls. This is in contrast to the escription of the data given by the authors: "With the exception of the lowest-dose TCDD group, all mice initiated with MNNG and treated with promoters had an increased number of papillomas and nodules" (p. 366). The document should be corrected. Significant skin tumor-promoting responses were seen with the congeners PCDF and HCDF, but there was a paradoxical dose-effect relationship throughout (e.g., lower doses produce higher responses), and this anomaly should also be pointed out. One aspect of this particular study is incorrectly represented in the EPA document. On page 6-23 of the EPA document, line 4-8, it is stated: Results (of the Hebert study) demonstrated that 2,3,4,7,8-CDF was 0.2 to 0.4 times as potent as TCDD and that 1,2,3,4,7,8-CDF was 0.08 to 0.16 times as potent as TCDD. These data suggest that the tumor-promoting potencies of structural analogues of TCDD, like the promotion of liver tumors, reflect relative binding properties of the Ah receptor. This statement is in contrast to what Hebert et al. (op cit.) said about the same data (p. 366): The lack of a clear dose-response makes it impossible to compare the relative potencies of the three compounds as promoters or to comment on the validity of the TEF approach for promotion as an endpoint. The potency estimates of 0.2 to 0.4 or 0.08 to 0.16, attributed mistakenly in the EPA document to promoting activity, are derived from a comparison of the relative changes in body weight and organ/body weight ratios (op cit. p. 372). This error should be corrected in the revised document. c) Lung tumors: The claim that TCDD promotes lung tumors is based on one experiment in which it was found that in ovariectornized rats, treated with DEN and given TCDD, lung tumors developed (four in 39 rats), whereas none were found in 37 intact animals treated with DEN and TCDD (Clark et al., 1991). No data on animals treated with DEN alone are available. The study was not published in the open literature (which should be noted in the document). The effect is statistically significant however, when analyzed with the appropriate test (uncorrected Chi-square). A second study on the promotion of lung tumors in mice by TCDD (Beebe et al., 1995) reported that TCDD enhances tumor multiplicity in the lungs of mice treated with DMN. This study needs to be discussed within the larger context of the murine lung tumor model being a representative system for tumor promotion. Although the EPA document provides a good over-all view on the studies done on liver and skin tumor promotion by TCDD, both a more critical analysis of the database and a more in-depth discussion of mechanisms underlying tumor promotion, should be added. Although some investigators believe that tumor promotion must result in neoplasms, the stage of promotion, in fact, only involves the development of preneoplastic lesions ranging from enzyme altered foci in rodent liver models to early papillomas in multistage epidermal carcinogenesis in the mouse. In model systems wherein the stage of promotion can be studied independent of the stage of progres sion, a number of characteristics of the stage and of the effects of promoting agents have been delineated (Yuspa and Poirier, 1988; Rahmsdorf and Herrlich, 1990; Pitot, 1993). Primary among these characteristics is the reversibility of the stage both at the level of gene expression and lesion growth. Promoting agents themselves typically exert their effects via receptor mechanisms and signal transduction. Promoting agents in various systems in vitro and in vivo inhibit programmed cell death and apoptosis (Schulte-Hermann et al., 1991; Magnuson et al., 1994; Wright et al., 1994). Although there are some other promoter hallmarks not yet established for TCDD, it conforms to all the characteristics of promoting agents noted above, and thus one is led to the conclusion that any carcinogenic effect of prolonged TCDD exposure is primarily, if not exclusively, the result of its action as a promoting agent. All of the evidence to date argues strongly that TCDD exerts its carcinogenic effect primarily through its effectiveness as a promoting agent, stimulating cell replication in a reversible manner and inhibiting apoptosis, both mechanisms presumably mediated through the Ah receptor and associated transduction mechanisms. TCDD is not a complete carcinogen and thus to avoid confusion should not be designated as such. Many structural congeners of TCDD appear to act in a similar manner to dioxin but there are as yet insufficient data to make any generalizations with respect to mechanisms of carcinogenesis or toxicity for all such structurally related chemicals. Finally, EPA needs to consider to what extent data on female rat liver foci should be used in modeling the tumorigenic activity of TCDD, be it as a promoting agent or as an incomplete carcinogen. The ratio of foci to tumors is far from being 1:1. Many foci may disappear when treatment is withdrawn; on the other hand, it is impossible that foci can grow to the size suggested by mathematical models, since one focus would occupy the entire liver, a fact pointed out in the document. Attempts to incorporate data into models and risk assessments will require a critical in-depth analysis of the biology of altered hepatocyte foci. 4.5.4 Characterization of Dioxin/Dioxin-like Compounds as Human Carcinogens Charge Question 13) Dioxin has been shown to produce malignancies in rats and mice of both sexes Although the epidemiological evidence linking dioxin exposure to the genesis of malignant neoplasms is limited, and does not offer compelling evidence of carcinogenicity to humans when taken by itself, this evidence is by no means inconsistent with such an effect. Almost all Members of the Committee therefore concur with the judgment that 2,3,7,8-TCDD, under some conditions of exposure, is likely to increase human cancer incidence. [Several Members contend that no epidemi-ological study has produced evidence that is widely accepted by the scientific community, including the IARC, as being convincing for the human carcinogenicity of dioxin.] The conclusion with respect to dioxin-like compounds is less firm. Since the information from animal studies is much less robust, and that from human studies often limited by uncertain ties about exposure, the judgment depends wholly on the similarity between the chemical effects of dioxin and those of its congeners, the dibenzofurans and other related compounds. For the congeners and dibenzofurans, enough evidence indicating similarity of biologic action exists to adopt the presumption that they too are likely to be carcinogenic to humans under some conditions (although in nearly all instances the doses required to produce the same incidence would be higher than those for dioxin). With respect to the polyhalogenated biphenyls which share only some of the physical characteristics and activities of dioxin, the degree of carcinogenicity has not been formally assessed in the present EPA document. However, at least one other authoritative body, the International Agency for Research on Cancer (IARC), has conducted such a formal assess ment and judged both polychlorinated (IARC, 1974; 1978a; 1987a) and polybrominated (IARC, 1978b; 1986; 1987b) biphenyls to be probable human carcinogens, both on the basis of animal studies as conclusive as those for dioxin, and, in the case of PBBs, with limited evidence from human studies. The Committee did not dispute the IARC judgement that PCBs and PBBs as likely to cause human cancer under some conditions of exposure. The Committee agrees that assignment of the dioxins, the PCBs, or PBBs to one of a mutually exclusive and collectively exhaustive set of carcinogenicity categories considerably oversimplifies the state of the science in most instances, possibly excepting those compounds for which there is an abundance of uniformly consistent evidence. However, prudent regulators must act and cannot base their regulation on ambiguous or inconsistent detail. They must make every effort to treat dangers of similar magnitude evenhandedly, even when there are limited management options. Although the level of exposure and the potency of the agent are measured on a meaningful continuous scale and can be incorporated into decisions on the basis of unambiguous continua, the degree of uncertainty re the human carcinogenicity of a compound is not measured in this manner, but is usually considered as a categorical term. It is desirable that consistent criteria for this inevitable categorization are employed. Under the proposed revisions to EPA s guidelines, there are essentially three alterna tive choices (unless "known to" is considered an alternative to "likely to"): a) likely to cause cancer under some conditions; b) not likely to cause cancer; and c) likelihood cannot be determined. In Chapter 9 of the reassessment document, the basis for a detailed multidimensional assessment is described and the various caveats are underlined, but a choice between these three alternatives (actually, between the first and the third--see below) is still required. Under the 1986 EPA cancer guidelines, three more levels of carcinogenic evidence, with mutually exclusive descriptive terms are provided, giving regulators more alternative choices. These choices include Group A -- human carcinogen; B1 --probable human carcino gen on the basis of limited information from human studies as well as animal studies; Group B2 -- probable human carcinogen on the basis of animal studies only; Group C -- possible human carcinogen; Group D -- not classifiable; and Group E -- evidence of non-carcinogenic ity for humans. In the case of dioxin, even if the additional alternatives were provided, virtually all of the Committee believes that the animal studies would be categorized as sufficient and the studies of humans as limited, providing for an overall categorization of B1, which would be expressed verbally as Probably Carcinogenic to humans with limited supporting information from human studies. PBBs and PCBs would receive ratings of B1 and B2, respectively. There is some merit in the provision of a slightly more detailed set of alternatives, but the provision of Group E, as well as of the proposed revised scheme category of not likely to cause cancer may be ill-advised. Since all mechanisms of carcinogenesis are not completely understood, and all sets of study conditions (species, dosage, co-carcinogens, etc.) cannot be foreseen, it seems unwise to suggest that evidences of non-carcinogenicity could ever be universally generalizable, at least as worded (with no reservations). Were one to employ a restrictive phrase such as under study conditions judged to replicate most recognized conditions of human exposure, the class would be more defensible, although one still would need to draw a difficult line based on the conditions of the negative studies. Please continue to Section 4.6. |
|||||||||||||||
|
The Cast | The Tragedy
| The
Comedy | The
Ending | News
| Library
| Seating
Plan | Top
|
Home |
CALL (812) 333-8888 FOR MORE INFORMATION OR E-MAIL info@copa.org | |||||||||||||||
The Coaltion Opposed to PCB
Ash in Monroe County, Inc. is a nonprofit organization. | |||||||||||||||